What causes measured sequencing depth variation

Question

[L. A. Liggett]

I am looking through a vcf file output by an amplicon sequencing protocol. This means that coverage of probed regions should be identical because the same number of fragments will cover just about every base pair in a probed region. Yet when I look across probed regions there will be some variation in the depth that is output. This isn't particularly large, perhaps of a coverage of 1000 there will be depth calls of 1005, 1003, 997 for different variants in the area.

So my question is how can there be variations in depth calls for areas that should be made up of the exact same number of supporting amplicons?

[GenoMax]

Is the variation you are seeing only for certain bases (should that not be explained by reads that were trimmed/have poor quality, at the ends of reads)?

[L. A. Liggett]

It is, so it shouldn't be explained by trimming or quality.